Background: Biomarkers may improve ascertainment of progression risk after radical prostatectomy (RP). We compared two validated post-RP classifiers: the Decipher genomic classifier (GC) and CAPRA-S (based on standard clinicopathologic parameters), to predict prostate cancer-specific mortality (CSM) in a contemporary cohort of RP patients. Methods: From a cohort of 1,010 RP patients treated from 2000-06, a case-cohort design was used to analyze a subset of 219 men with one or more high risk features and available paraffin-embedded tissue. Median follow-up was 6 years. The GC, derived from expression levels of 22 biomarkers and dichotomized to denote low- and high-risk, and CAPRA-S, calculated from preoperative PSA and pathologic grade and staging variables, scores were determined. The scores were evaluated individually and in combination using concordance index, decision curve, (DC), re-classification, and Cox analyses for prediction of CSM. Results: 212 men had full data available to calculate the CAPRA-S; 27 experienced CSM. The c-index for GC (0.78) and CAPRA-S (0.76) were similar, although GC showed improved calibration and higher net-benefit on DC analysis. In 103 patients with high-risk CAPRA-S scores (≥6), GC scores were likewise high-risk for 49, among whom 19 had CSM events. The other 54 men were reclassified as low-risk by GC; among these only 1 CSM event was observed. In multivariable Cox analysis both GC and CAPRA-S were independently prognostic of CSM, with hazard ratios of 1.62 (p<0.001) and 1.22 (p=0.01), respectively for unit score increases. A combined model defined based on the Cox model as (0.20*CAPRA-S + 5.68*GC) was more accurate than either score alone (p<0.001 by likelihood ratio test). DC analysis indicated greater net benefit for the combined model than for either score alone. Conclusions: In men treated with RP at high risk of recurrence based on clinical and pathologic variables, both GC and CAPRA-S were significant predictors of CSM. Notably, GC was able to 'down-risk' >50% of men stratified to high risk based on CAPRA-S alone. Thus the GC provides independent prognostic information, and a model integrating GC and CAPRA-S may further improve the prediction of lethal prostate cancer.